The Deutsche Bank QIS Research Team has launched a new podcast series titled 'Stochastic Conversations', designed to provide greater insight into the backgrounds and current perspectives of their quant researchers. The series explores various market drivers through the lens of individual expertise, which may offer institutional traders unique perspectives on assessing market risks and opportunities. Given the recent volatility in cash equity markets linked to macroeconomic factors, this series could hold relevant insights for navigating current trading strategies. Per the full note source, the discussions will encompass underlying scientific methods and influences on trade decisions that could impact portfolio allocations.
What the desk is arguing
The thesis posited by the Deutsche Bank QIS Research Team highlights the intersection of scientific inquiry and financial market analysis in their new podcast series. As the team explores the backgrounds of individual analysts, the desk underscores the relevance of academic rigor in understanding market dynamics and the quantifiable aspects of trading. A specific focus will be placed on how non-financial backgrounds can enrich perspectives in finance-driven environments.
Evidence can be drawn from Gianpaolo Tomasi's background in brain imaging and data science, illustrating the potential for unconventional entry points to yield superior analytical skills in finance. This type of interdisciplinary approach may resonate particularly well in the current market environment, where cognitive frameworks often shape traders' responses to data anomalies.
The alternative view could suggest that traditional financial backgrounds predominate in achieving success in quantitative finance roles, thus underestimating the value of diverse sectors. However, the depth of analysis presented by non-traditional players may challenge this narrative.
01The new podcast series links academic and quantitative disciplines to market analysis.
02Individual backgrounds can enhance predictability in market behaviors.
03Non-traditional pathways in finance lead to diverse analytical perspectives.
04Market dynamics may be more comprehensively understood through interdisciplinary approaches.
Market implications
Watch for how advancements in quantitative techniques are employed in trading strategies, particularly given the current focus on cash equity market drivers affecting broader asset classes. Any shifts in quantitative interpretations may directly influence positioning in equity indices and volatility metrics.
Risks to this view
Should the podcast highlight unsuccessful analytical strategies or misinterpretations of market data attributed to non-conventional perspectives, it could lead to skepticism surrounding such interdisciplinary frameworks. Additionally, significant shifts in macroeconomic indicators could invalidate this approach, as traditional finance practices regain dominance in market interpretation.
Hello everybody and welcome to our latest Stochastic Conversations podcast. The podcast here at Deutsche where you get to meet the individual quant researchers in our team. My name is Caio and I will be your host today.
Today I'm pleased here to have a conversation with Gianpaolo Tomasi who leads our work in quant cash equities. Hello everybody, great to be here. So Gianpaolo, let's get down to it.
You have a science background, right? You're formerly a scientist and you come from Italy, as I can see from the accent. What motivated you to A, leave Italy and then B, pursue a career in academia?
Yeah, so both of them were quite random unplanned events. I was never planning on leaving Italy, I was very happy there, it just so happened that after the end of my PhD, I did not have any opportunity to stay in academia and got a very good offer from Yale and so I decided to move, it was a hard decision but I went for it. And then after several years in academia, I kind of realized that maybe it was not the right choice for me, it was not as I would expect it to be and I started looking for jobs but admittedly, I didn't have any interest or knowledge about finance, I didn't know what S&P was back then.
What were you studying, what were you doing in academia? I was doing brain imaging from a data science perspective but it's called PET, positron emission tomography, you study how the brain absorbs certain substances and it's very useful for diagnosis of certain conditions, cancer, it's very, very interesting. Yeah, no, I was just random, I actually interviewed for, back then, for Google, for a data science position with Tesco and for an edge fund and of course, my career would have been very different and I got in one of the other jobs but I got the job.
So, you were used to looking at brain images and then you decided to look at images of squiggly lines and charts and time series of prices going up and down, is that the idea? Quite a big change indeed, it was a bit of a shock. So, you've, I mean, I can see here that you worked for five years on the buy side, particularly in the hedge fund industry before joining Deutsche in 2018.
So can you tell us a bit about what were the key differences between buy side and sell side when it comes to being a quant researcher? So just tell us a bit about what was your day-to-day job when you were a quant researcher on the buy side and then how did that change as you came to the sell side? So there are a lot of similarities in the sense that in both positions, most of my focus is to come up with new ideas and new signals, new way to kind of create systematic equities, equity strategies.
At the same time, there are very big differences. For instance, at the buy side, everything's focused in the improvement of your current signal. Here, as you know, sometimes we just ask ourselves interesting research questions.
They might or might not have the right implication on the signal. So there's more like research just for the sake of research, which is less probably common on the buy side. That's one thing.
And of course, there's many other differences. Here we interact a lot with clients. At the buy side, it was mainly interaction with your colleagues.
And the focus essentially is P&L. It's very interesting. It's a faster environment.
It's, in my experience, way more stressful because you can be successful or unsuccessful and then it affects your day-to-day. Here it's, yeah, there are a lot of similarities. But I think, yeah, at the same time, there are very different types of jobs and positions and pros and cons for everything else in life.
So what was your, I mean, tell us a bit about your daily routine when you were on the buy side and how does that compare with your daily routine on the sell side? What are the key differences there? I see that you cycle to work, for example, right?
That does not change. I was cycling to work. You've always cycled to work.
Yeah, I mean, I was in Oxford, so you have to cycle in Oxford. It's mandatory, even in London. So I think the key difference, especially for a peer where I was managing a small portfolio, was the first thing you do in the morning there is you look at the P&L that you accrued overnight and that can already make or break your day.
If you were up 20 bps, it's a great day, 30 bps. If not, the start of the day doesn't start very well. In terms of day-to-day, this in my case was 100% research, as in, you know, read papers, think of code, talk to your colleagues.
Here the job at Deutsche has a lot of other components, such as interaction with clients, interaction with colleagues of our team, interaction with colleagues of our other teams, interactions with data vendors. So it's a lot more variability, higher standard deviation here in terms of the activities, whereas there was just, you know, signal research, portfolio construction and all day long, essentially. So what are the most and perhaps the least interesting aspects of your job?
So how do you classify them? Yes, I do love the fact that this type of career is, you know, the learning process is never-ending. If you like to learn, even your field of expertise, there are always new things to learn from reading papers, to talk to colleagues, to test new ideas.
And then, of course, because in our team we try to be as, you know, to keep the breadth quite big, that we need to learn to know about what our colleagues are doing, which means study new asset classes, study new ideas. So the learning process is continuous and I definitely do enjoy that because you never get bored. On the flip side, like in other jobs, there are some more kind of mundane aspects to it, more somewhat meant to be done that you know quite well, because for you it's even worse.
But that's true for every job. Ultimately, everything is a balance. And if you're lucky enough that the nice interesting bits are, you know, in terms of time are much more than the less interesting bits, then I'm happy and lucky.
You've been traveling a lot, right? As far as I recall, over the course of this year, you've been in the US for quite a number of weeks and then you were in other regions as well. How did you get used to the travel routine?
How does that compare with when you were on the buy side? Is it completely different? It is.
When I was at the buy side, I was a bit more junior, so I was not really client facing. Here it's a bit more, yeah, definitely interaction with clients are central. Ultimately, that's why in a way we work for our clients.
I think in general, very important things in that space are communication skills, probably more so here than at the buy side, to be able to deliver complex concepts to an audience which is sometimes very highly sophisticated, other times maybe less so. So you need to be able to adapt to the tone of the conversation and to be able to read the room in a way. So definitely soft skills are probably more important here than they were at the buy side.
That's for client interaction and try to understand what's the right thing to say and what not to say. You know this way better than me, I've been doing this for 20 years. But I remember you get to enjoy some of the travel too, right?
I do recall you having a very casual walk at the Tenderloin District in San Francisco with, you know, while others of us were a bit scared of the neighborhood that we were walking, but you were very chilled, very, I suppose very Italian about it. So you get to enjoy some of your travels as well. I do.
You promised like a proper zombie walk. I was a bit disappointed because there weren't really any zombies, but yeah, yeah, it was a nice trip. I got to see a lot of places, although very quickly, and you got to experience many things.
Yeah, one go. Okay. So now let's zoom into work and what you do as the person that leads our research in cash equities within the Quant Research team.
So can you tell us a bit about this concept that you often talk to clients about, which is the artisan approach to Quant Equity Investing versus the kitchen sink approach to Quant Equity Investing? I know that you look at both of these styles, and so just tell us a bit about this. Sure.
In both cases, the idea is to come up with a multifactorial strategy where you profit ideally from the exposure to different styles, like value, quality, momentum. In the first case, however, we typically start with a relatively low number of factors. In our case, it's four.
It could be five, six, typically less than 10. We spend a lot of time on each of them. There's a lot of research.
In our case, we have plenty of, in the past and even today, with fundamental experts that look at any single signal with a lot of depth. We debate internally about the details of the construction. And so there's a lot of time spent on the individual building blocks.
And then when it comes to aggregation, typically we take a very, if you want, agnostic approach. We believe equally in all these factors. In our case, it's four, momentum, value, quality, low beta.
And we typically take either inverse volatility weighting approach or equal risk contribution approach. We believe that we equally believe in all of them. We have no view about what's going to work in the next period.
So we take an agnostic view. So the emphasis is mainly on the signal research and less on, if you want, portfolio construction. That's something that we have.
We work a lot with products and some clients like it because it's quite simple. It's quite easy to explain. It's backed by a lot of academic research.
So there's an element of simplicity that really resonates well with some clients. But then there's also the kitchen sink approach, right, which is quite distinctly different from the autism approach. How would you categorize that?
So that's fundamentally different because we start with a very big set of initial signals slash alpha slash features. But we do not spend a lot of time on each of them. Typically, we take them from the academic literature.
And typically, in our case, there are one line, two lines of Python code for each of them. So it's ratios or some kind of delta. So nothing really complicated.
But then the emphasis is on the aggregation, how do you optimally put them all together. And here, the challenges are that, number one, because you have so many of them, you need to model their interactions, their correlations. Obviously, in our case, you can have variation of each other.
The signal could be a slight variation of each other. So their correlation could be extremely high. And you really need to model that factor.
And then the second component you need to model is the fact that you can't really believe that each of these under the feature have the same long-term forecasting power. Because it can't simply be true. And you need to find a way to properly weigh this feature such that, ideally, in the end, you get with the best optimal portfolio.
So to account for their expected future forecasting power. So definitely, it's more on the aggregation slash portfolio construction and a bit less on the signal. So they're complementary.
And sometimes, of course, it's performance. But also, it's a matter of personal preference, depending on what the client prefers. Thanks.
So let's zoom into this kitchen sink approach a bit more, and the idea of feature engineering and signal aggregation and so on. I wanted to talk a bit about N-LASER, which is this model, this strategy that the bank has been publishing on for the past 15 years. And I believe six and a half years ago, you published on it as well.
And then more recently, too, talking about some innovations that we could bring into the strategy. So tell us a bit about what N-LASER is and what's unique about it. Yeah.
So as I say, it's a topic where the Deutsche Bank has been engaged since 2012. And we wrote, as a bank, four proper research papers. And more recently, we wrote an overview of the performance.
So N-LASER belongs to the second class of approaches, which described earlier, the kitchen sink approach, where we start with a huge number of factors, in our case, more than 100. And they're quite traditional. They're taken from the academic literature.
In our case, they are low turnover. We intentionally did not include, back in the days, anything really fast. So we have no mirror version, nothing news-based, which simplifies the problem, because at least the turnover of these alphas is quite homogeneous.
So they're very similar to each other. And then, as discussed before, the problem is the aggregation, and it's an algorithm which is inspired by Adaboost, where we model the... It's essentially, it's a diversified factor momentum, where we give a higher weight to the factors which perform well in our training windows.
But also, we account for their correlation, and we enforce diversification while we build the portfolio, because we are aware that we should never be overexposed to any style factor, because we don't really know what's going to work in the next month. So we need to try to be as balanced as possible. And then, I think one thing I really like, especially I've never seen anywhere, is the idea of training the model in different training windows in the past, and then essentially averaging the four portfolios created by the four different training windows.
Here, the idea is that we know that by aggregating signals of similar predictive power, but not identical, the average tends to be better than the individual components. That's a very well-accepted principle of machine learning, called assembling or stacking. And quite interesting, we have seen in the five and a half years of AutoSample history, that's true in our case as well.
So there's a lot of thought behind this, and spent a lot of time as a team on this project, and it's been quite successful. As you know, we have a live performance for the Sharper 1.5 in five years and a half. So we want to even expand it to different universities.
It's a quite exciting project, it's very, very good. So thanks for that. So moving on to basically the future, quant investing has changed significantly over the past decade.
What do you think are the most important trends that are shaping equity factor investing today, and how do you think we should position for those? Yes, so definitely there's more widespread use of machine learning for combining the factors. Although, especially at the frequency we typically look at, in our case, which are what you call low turnover portfolios, a lot of care should always be taken in order to avoid overfitting, because there's so many very complex three-base neural network-based approaches that, in the case of low-frequency portfolios, there's a very high risk of overfitting.
So that's one big trend, because it's not only us, there's many papers on this topic. And then, of course, the second big topic of alternative data. Nowadays, there's been an explosion of new data sets, and here the challenge is simply also to select what to test, because it's not possible to test these thousands of data sets provided by hundreds of data vendors, each of which is telling us that their data is great.
So even deciding how you want to spend the next couple of months is crucial. And as you know, we're trying to move towards, well, first of all, we like the idea of having a third-party data vendor that kind of gives us an unbiased opinion, so this alternative data, you know, because essentially it's impossible to do it by yourself. And then number two, as you know, we're trying to move with our colleague Y in kind of automatic testing of all this.
I mean, it's not for now, but for the future, in a kind of agentic framework where a lot of the backtest is done automatically, so we can test a high number of these data sets with a low effort. It's not that we are very excited, we're not there yet, but we're definitely moving in the right direction. Okay.
Which brings me to a question. If you were building a new quant equity research team from scratch, so if you're starting a new team and you're hiring people, right, in this new age where agentic research is becoming increasingly widely used, what kind of skills would you be looking for in your hires for a quant research team, for a quant research role? Yeah, that's in general a comment about what are most important skills to be successful in this career.
Definitely quantitative skills are essential. And although AI is changing and is going to change even more, so coding, I think coding is still a very important skill because I think we agree coding will change, but still it's necessary to be able to know the principle, the basics. So definitely coding and basic quantitative skills are the base.
Then as I mentioned earlier, especially it's not only sales side, but also by side explainability skills. You need to be able to talk to your colleagues, explain what you're thinking, explain very complex concepts in a way that everybody can understand. And even more so for us, because we have a very heterogeneous client base, so definitely crucial.
And then of course there's third element, which is soft skills. Like, you know, ultimately this is a people job, we interact with colleagues, we interact with clients, we interact with members of different teams. And it's very important to be able to do, you know, appropriately, like to be able to, you know, interact with the peers.
And so it's not only about, you know, being the best quant researcher, it's we're still working in a big organization where, you know, you need to be able to work with different people. So yeah, it's a combo of all these different things. So that's what makes also the job so interesting.
Okay. Now, let me move on to talking about a more serious topic. Coffee.
Coffee. I know it's a very important topic to you. It is a very important topic, yes.
And I see you sometimes drinking from a Nescafé cup of coffee. And I mean, I'm sure it's just, it's just the label of the cup. You don't actually drink Nescafé coffee, right?
I mean, you're Italian and I mean, you guys take coffee very seriously. So can you elaborate a bit and just just clarify that you don't actually drink Nescafé, you drink something more? No, unfortunately, you're wrong.
I do admit my sins. I do drink Nescafé. Like I didn't until I moved to the US and started drinking this big, you know, Starbucks style coffee.
Before it was simply classical macchiato. And so now and then you get used to this, you know, big amount of hot fluid get inside your body. It's very un-Italian.
That's very un-Italian. I know that's a point of contention with you. Do you drink cappuccino after 11 o'clock?
As you know, that's forbidden in Italy. It's an unspoken rule that helps us identify the tourists from the proper Italian. So as you know, you don't do that at 10.30.
So you don't. 10.30 is fine. But Nescafé is okay. Yeah, Nescafé is okay.
A funny story about this, when I go back to Italy, which is quite often, I do go back to the Italian proper coffee culture, which is more macchiato and that's it. So I'm a personal habit and I adapt to my surroundings. But in London, yeah, it's Nescafé and I'm proud of it.
I suppose you like caffeine. Is that the idea? Yeah.
Yeah. I like caffeine and it's a nice, you know, treat yourself. So that's how it is.
So, you know, talking about Italy and what is a commonly held belief that people have about Italy that you think is just completely wrong? Yeah. So food is great, art is great, my things are great.
The lakes are great. The lakes? The lakes are great.
The lakes are great. They're very nice. I'm going to go to Lake Garda very soon.
Garda? Okay. In my case, the weather is maybe not as...
I'm from the north and, yeah, it can get really, really cold during the winter. It's not that different from London during six, seven months a year. So definitely it wasn't a shock in terms of the weather.
We have very long, nice summers, but, you know, especially after this year in London, I don't think I can... After all the heat waves, I think I feel more... I never suffer because of the weather.
I suffer a bit because of the tomatoes don't taste as nice, but... The tomatoes don't taste as nice? Tomatoes here don't taste as nice as in Italy.
You definitely can tell that the broccoli and the tomatoes and the grapes have a different flavor. Okay. Okay.
I didn't know you were that picky about tomatoes. Okay. Fine.
And you, I mean, you've lived in a number of countries and you worked in different environments. What kind of place feels like home, most like home for you today? Yeah.
So I lived for two years in the US and then I've been here in the UK for 16 now. I'm married and have house. Children are born here, but still...
So the textbook answer, yeah, you know, every place has a different role in my life. Ultimately, I still feel at home only when I go back to Italy, just, you know, I live almost 30 years there and so I have my family, my friends, and it's always great to go back there. But, you know, I still, you know, I like living in London.
I just, you know, I don't support England as a football team. I don't, you know, I can't really... For me, football team is Italy.
It's Italy, right? Yeah, yeah, yeah. So that's a bit disappointing.
The latest World Cup. For Sazuri, how do you call it? Yeah, for Sazuri.
For Sazuri. Not for a long time now, but yeah, that's... I suppose when you live so many years of your life in a place that doesn't matter.
Yeah, so that's it. So let me ask a final question. I mean, I'm sure you have hobbies, right?
As every eclectic scientist has hobbies too. What is the most unusual hobby or interest that you have that people that don't know you would... Or even some that who do know you would be surprised to learn about?
Yeah, I don't know what I wrote in that answer, but definitely I play as an Italian, I play football all my life for 15 years, which I loved. And then I got into running and I even ran a marathon when I was younger. And so definitely maybe my colleagues don't know much of my past running life.
Nothing's professional. I was just running. Not so much time nowadays because of work, family issues, not issues, family commitments.
But definitely maybe this side of me is unknown. I don't speak as much. But you still do your cycling?
I still do my cycling. I have a very nice yellow helmet and it's my way to kind of exercise. And I don't like the tube, especially in the morning.
The line is horrific. So I rather have my yellow helmet and cycle from bottom to the city. You can see me.
Yes, yes. Yes, you keep reiterating the color of the helmet. So I appreciate that.
I do. Right. Okay.
Well, thank you very much, Gianpaolo. Thank you very much for the audience, to the audience as well. If you have any questions about Gianpaolo or what we do in the PONT Research team here at Deutsche, feel free to send us an email or to reach out to your sales coverage directly.
This has been the latest edition of our Stochastic Conversations podcast. My name is Caio. I was your host today and we look forward to speaking to you in the next one.
This podcast has been produced by Deutsche Bank and may contain research as defined in Method 2. The information discussed is believed to be reliable and has been obtained from public sources believed to be reliable, although Deutsche Bank makes no representation as to its accuracy or completeness. Opinions, estimates and projections discussed constitute the current judgment of the speaker at the time of recording.
They do not necessarily reflect the opinions of Deutsche Bank and are subject to change without notice. For further important information, please visit research.db.com.